DNA-Unwinding Dynamics of Escherichia coli UvrD Lacking the C-Terminal 40 Amino Acids

Abstract

The E. coli UvrD protein is a nonhexameric DNA helicase that belongs to superfamily I and plays a crucial role in both nucleotide excision repair and methyl-directed mismatch repair. Previous data suggested that wild-type UvrD has optimal activity in its oligomeric form. However, crystal structures of the UvrD-DNA complex were only resolved for monomeric UvrD, using a UvrD mutant lacking the C-terminal 40 amino acids (UvrDΔ40C). However, biochemical findings performed using UvrDΔ40C indicated that this mutant failed to dimerize, although its DNA-unwinding activity was comparable to that of wild-type UvrD. Although the C-terminus plays essential roles in nucleic acid binding for many proteins with helicase and dimerization activities, the exact function of the C-terminus is poorly understood. Thus, to understand the function of the C-terminal amino acids of UvrD, we performed single-molecule direct visualization. Photobleaching of dye-labeled UvrDΔ40C molecules revealed that two or three UvrDΔ40C molecules could bind simultaneously to an 18-bp double-stranded DNA with a 20-nucleotide, 3′ single-stranded DNA tail in the absence of ATP. Simultaneous visualization of association/dissociation of the mutant with/from DNA and the DNA-unwinding dynamics of the mutant in the presence of ATP demonstrated that, as with wild-type UvrD, two or three UvrDΔ40C molecules were primarily responsible for DNA unwinding. The determined association/dissociation rate constants for the second bound monomer were ∼2.5-fold larger than that of wild-type UvrD. The involvement of multiple UvrDΔ40C molecules in DNA unwinding was also observed under a physiological salt concentration (200 mM NaCl). These results suggest that multiple UvrDΔ40C molecules, which may form an oligomer, play an active role in DNA unwinding in vivo and that deleting the C-terminal 40 residues altered the interaction of the second UvrD monomer with DNA without affecting the interaction with the first bound UvrD monomer.

Significance

E. coli UvrD is a superfamily 1, nonhexameric DNA helicase required for DNA-repair mechanisms. Previous studies have suggested that this helicase has optimal activity in its oligomeric form. Nevertheless, a conflicting monomer model was proposed using a UvrD mutant lacking the C-terminal 40 amino acids (UvrDΔ40C), suggesting that UvrD cannot dimerize, and the UvrD monomer is required for DNA unwinding. Here, single-molecule direct visualization of UvrDΔ40C revealed that two or three UvrDΔ40C molecules were simultaneously involved in DNA unwinding, possibly in an oligomeric form, similar to that with wild-type UvrD. Thus, this study addresses the role of C-terminal amino acids in the helicase mechanism and identified its functional significance in nucleic acid binding in proteins with helicase and dimerization activities.

Introduction

Previous structural and biophysical studies have shown that protein oligomerization is a key factor in the regulation of proteins (1). A search of the Brenda enzyme database (https://www.brenda-enzymes.org/), the most comprehensive enzyme repository, indicates that only about one-fourth of enzymes are monomers.

Helicases are enzymes that unwind nucleic acids and play major functional roles in genome maintenance, such as DNA replication, repair, and recombination, without exception. One class of helicases, namely those in superfamilies (SFs) 3–6, can function as hexameric ring structures that can encircle DNA (2,3), whereas members of the other classes, which includes SF1 and SF2 helicases, function in a nonhexameric form (4,5). Data from recent studies have suggested that some nonhexameric helicases also function in oligomeric forms (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20).

Among the SF1 helicases, the tertiary structures of the E. coli UvrD and Rep helicases and the Bacillus stearothermophilus PcrA helicase have been resolved by x-ray crystallography (6,21, 22, 23). These structures provided clear evidence that these helicases share high homology in their primary structures (∼40%) and contain four structural domains (1A, 1B, 2A, and 2B). Based on the crystal structures, it was proposed that the 2B subdomain plays a catalytic role in DNA unwinding (21,23,24). Conformational changes in the 2B subdomain are induced by helicase binding to double-stranded DNA (dsDNA), which changes the helicase structures from opened to closed conformations.

Despite their high structural homologies, two conflicting models have been proposed to explain DNA unwinding by nonhexameric helicases. The dimeric helicase model has been proposed for SF1 UvrD (7, 8, 9, 10, 11, 12), Rep (6,13,14), PcrA (15) helicases, the SF2 NS3 helicase (16), and DEAD-box RNA helicases CsdA (17), CshA (18), RhlB (19), and Hera (20). In addition, the monomeric helicase model was proposed for SF1 E. coli UvrD (21,25), PcrA (23,26), and SF2 hepatitis C viral NS3 RNA helicases (27).

E. coli UvrD (720 aa, molecular mass = 82 kDa) is an SF1 DNA helicase that plays a vital role in both nucleotide excision repair and methyl-directed mismatch repair (28). UvrD contains four structural domains, namely 1A (aa 1–89 and 215–280), 1B (aa 90–214), 2A (aa 281–377 and 551–647), and 2B (aa 378–550) (21), and has an unstructured C-terminal region (aa 645–720) (21,29). Using energy derived from ATP hydrolysis, the UvrD helicase unwinds duplex DNA starting from the 3′-end single-stranded DNA (ssDNA) tail at a gap or a nick. Data from previous biochemical studies have suggested that this helicase has optimal activity in its oligomeric form (7); this hypothesis is supported by a single-molecule DNA-manipulation study conducted using magnetic tweezers (8) and was supported by direct visualization of single UvrD molecules (9, 10, 11). Data from another recent study showed that UvrD dimerization, the binding of a second UvrD to a first bound UvrD, shifted the 2B conformation of the first bound UvrD to a more closed state, resulting in the activation of helicase activity (12).

In contrast, the conflicting monomeric helicase model was proposed based on crystal structures of the UvrD-DNA complex resolved only for monomeric UvrD (21) and a study involving genetic and biochemical experiments (25,30). In these studies, a UvrD mutant lacking the C-terminal 40 amino acids (UvrDΔ40C) was used. By performing genetic-complementation assays, Mechanic et al. found that UvrDΔ40C was fully active in terms of methyl-directed mismatch repair and nucleotide excision repair. They also found that UvrDΔ40C was slightly defective at ssDNA binding, but was fully competent at DNA unwinding, compared with wild-type UvrD. Together with their additional results showing that UvrDΔ40C could be eluted as a single peak by size-exclusion chromatography and that analytical sedimentation equilibrium ultracentrifugation experiments showed no increase in the apparent molecular mass of UvrDΔ40C with increasing protein concentrations, they concluded that UvrDΔ40C was unable to dimerize and proposed the monomeric model.

The effects of deleting 73, 102, or 107 C-terminal amino acids from UvrD (creating UvrDΔ73C (UvrD_1–647), UvrDΔ102C, or UvrDΔ107C) were also reported (25,29,30). UvrDΔ102C and UvrDΔ107C failed to substitute for the wild-type protein in methyl-directed mismatch repair and nucleotide excision repair. Concerning the ssDNA-binding affinity, UvrDΔ73C (UvrD_1–647) showed slightly reduced affinity, whereas UvrDΔ102C showed a marked reduction in affinity. Consistent with their ssDNA-binding affinity data, UvrDΔ73C (UvrD_1–647) retained ATPase and DNA-unwinding abilities, whereas those abilities were lost with UvrDΔ102C. With the UvrDΔ102C mutant, all C-terminal amino acids and part of domain 2A, which has a conserved DNA-binding motif, were deleted (21). Thus, the loss of DNA binding, ATPase, and DNA-unwinding activity for UvrDΔ102C was likely due to disruption of domain 2A rather than the complete loss of the unstructured C-terminal region.

The C-terminus plays essential roles in nucleic acid binding for many proteins with helicase and dimerization activities. However, the function of C-terminal amino acids in dimerization and DNA unwinding is poorly understood. Thus, to understand the interrelationships between the C-terminal amino acids, oligomerization, and DNA-unwinding activity, single-molecule direct visualization (9) was performed for UvrDΔ40C, which was used for biochemical, genetic, and x-ray crystallography studies that supported the proposed the monomeric model.

In this study, a photobleaching step analysis (9,31,32) was performed to quantify helicase binding to DNA in the absence of ATP, revealing that multiple UvrDΔ40C molecules bound to an 18-bp dsDNA with a 20-nt 3′ ssDNA tail simultaneously. Binding of UvrDΔ40C molecules was also observed on a DNA substrate with a shorter 3′ ssDNA tail (18-bp dsDNA with a 12-nt 3′ ssDNA), which was considered to accommodate fewer UvrDΔ40C molecules. Then, to determine whether multiple UvrDΔ40C molecules were involved in DNA unwinding in the presence of ATP, simultaneous single-molecule visualization studies were performed for DNA-unwinding events driven by the helicase and of association/dissociation events between the helicase and DNA (9). UvrDΔ40C completely unwound the DNA after two or more of the helicases were bound to it, as observed with wild-type UvrD. The association of multiple UvrDΔ40C molecules just before completion of the DNA-unwinding process was also observed on 18-bp dsDNA with a 12-nt 3′ ssDNA, which implies that these proteins might have interacted with each other and formed an oligomer. In addition, both the determined dissociation and association rates increased as the number of UvrDΔ40C molecules bound to DNA increased, as was observed with wild-type UvrD. The association/dissociation rate constants of the first bound UvrDΔ40C monomer with/from DNA were similar to those for wild-type UvrD, whereas the corresponding rate constants for the second bound UvrDΔ40C monomer were ∼2.5-fold larger than that for wild-type UvrD. These results strongly suggest that multiple UvrDΔ40C molecules are involved in DNA unwinding and that deleting 40 residues from the C-terminal end altered the interaction of the second UvrD monomer with DNA but not the first UvrD. The involvement of multiple UvrDΔ40C molecules in unwinding DNA was also observed under a physiological salt concentration (200 mM NaCl), which showed that the DNA-unwinding dynamics observed in this study are relevant to their in vivo function. Thus, quantifying the helicase association/dissociation rate constants by single-molecule direct visualization shed light onto interrelationships between the C-terminal amino acids, oligomerization, and unwinding activity of the nonhexameric helicase UvrD.

Materials and Methods

Detailed information on preparing DNA substrates and the UvrD protein and single-molecule imaging assays are available in the Supporting Material.

Buffers

The buffers used in this study were prepared with reagent-grade chemicals and deionized water, which were obtained using a Milli-Q system (Merck Millipore, Darmstadt, Germany).

DNA substrate used for single-molecule imaging assays

The duplex DNA substrates used for single-molecule imaging assays were prepared through the hybridization of 5′-Cy3-TGGCGACGGCAGCGAGGC-(T₂₀)-3′ or 5′-Cy3-TGGCGACGGCAGCGAGGC-(T₁₂)-3′ with 5′-GCCTCGCTGCCGTCGCCA-biotin-3′.

UvrD protein

UvrDΔ40C with Cys-to-Ala mutations was expressed, purified, and labeled with Cy5 maleimide (PA25001; GE Healthcare Japan, Tokyo, Japan) at a UvrD:dye molar ratio of 1:3 in 500 mM NaCl, 10% (v/v) glycerol, and 20 mM HEPES (pH 7.0) for 20 h at 4°C (9). After the unlabeled dyes were removed, the labeling ratio was determined by ultraviolet-visible spectrophotometry. The labeled proteins were aliquoted, quickly frozen in liquid nitrogen, and stored at −80°C until use.

Single-molecule imaging assays

All single-molecule experiments reported here were performed at 25°C using one of two types of flow cells: a nail-polish-sealed flow cell for photobleaching-step analysis in the absence of nucleotides or in the presence of adenosine 5′-(γ-thio)triphosphate (ATPγS) (Figs. 1 A and 2 A) or a flow cell sealed with double-sided tape for DNA-unwinding assays conducted in the presence of ATP (Fig. 4 A; (9)). Both types of flow cells employed objective-type total internal reflection fluorescence microscopy.

Figure 1.

Single-molecule visualization of Cy5-UvrDΔ40C bound to a duplex DNA substrate with a 3′ ssDNA tail in the absence of nucleotides. (A) Shown is a schematic representation of the experimental design. First, a Cy3-labeled 18-bp dsDNA with a 20-nt 3′ ssDNA tail and biotin at one end was immobilized on a PEGylated quartz slide via streptavidin-biotin interactions. Then, 50 μL of 2.0 nM Cy5-UvrDΔ40C in buffer U (6 mM NaCl, 2.5 mM MgCl₂, 10% (v/v) glycerol, and 25 mM Tris-HCl (pH 7.5)) with an oxygen-scavenger system was infused. After 5 min, single-molecule visualization of Cy3-DNA and Cy5-UvrDΔ40C was performed by objective-type total internal fluorescence microscopy, in which Cy3 and Cy5 were excited at 532 and 637 nm, respectively. (B and C) Shown are time courses of the fluorescence intensity (average of three moving frames, 0.1 s each) of the fluorescent Cy5-UvrDΔ40C spots. The photobleaching processes indicated by the arrowheads occurred in one (B) or two (C) steps. To see this figure in color, go online.

Figure 2.

Single-molecule visualization of Cy5-UvrDΔ40C bound to a duplex DNA substrate with a 3′ ssDNA tail in the presence of 1 mM ATPγS. (A) Shown is a schematic representation of the experimental design. Immobilization of the Cy3-labeled 18-bp dsDNA with a 20-nt 3′ ssDNA tail on a PEGylated quartz slide was performed as described in Fig. 1A. After DNA immobilization, 50 μL of 2.0 nM Cy5-UvrDΔ40C in buffer U with 1 mM ATPγS and an oxygen-scavenger system was infused. After 5 min, single-molecule visualization of Cy3-DNA and Cy5-UvrDΔ40C was performed as described in Fig. 1A. (B–D) Shown are time courses of the fluorescence intensity (average of three moving frames, 0.1 s each) of the fluorescent Cy5-UvrDΔ40C spots. The photobleaching processes indicated by the arrowheads occurred in one (B), two (C), or three (D) steps. To see this figure in color, go online.

Figure 4.

Simultaneous single-molecule visualization of DNA unwinding driven by Cy5-UvrDΔ40C and the association/dissociation events between the helicase and DNA in the presence of 1 mM ATP. (A) Shown is a schematic representation of the experimental design. Lasers for exciting Cy3 at 532 nm and Cy5 at 637 nm were incident on the sample plane with objective-type total internal fluorescence microscopy. Single-molecule fluorescence signals from Cy3-DNA and Cy5-UvrDΔ40C were simultaneously imaged using a dual-view apparatus. For further explanation of the experimental setup, see the text. Traces in which the Cy3 and Cy5 fluorescence intensities decreased almost simultaneously to their respective background levels were analyzed. Because of the criteria for selecting such traces, the effect of Cy3 and Cy5 photobleaching was negligible though indistinguishable. (B) Shown are typical time traces of the Cy3 and Cy5 fluorescence intensities (F. I.), in which the Cy5 fluorescence intensity increased in a two-step manner just before DNA unwinding, and the Cy3 fluorescence intensity decreased to its background level. (C) Shown are time traces of the Cy3 and Cy5 fluorescence intensities, in which the Cy5 fluorescence intensity changed in a three-step manner just before the DNA-unwinding process; specifically, the Cy5 fluorescence increased in two steps but decreased in three steps. The three-step fluorescence decrease is indicated with arrows. (D) Shown are typical time traces of the Cy3 and Cy5 fluorescence intensities, in which the Cy5 fluorescence intensity increased in a single step just before the DNA-unwinding process. (E) Shown is the distribution of the dwell time (τ₂) between the appearance of the second step and completion of the DNA-unwinding process. The mean dwell time, which was obtained using a single exponential fit, was 2.1 ± 0.1 s. (F) Shown is the experimentally obtained distribution of the number of step changes in the Cy5 fluorescence just before or after DNA unwinding. (G) Shown are the theoretical distributions of the number of steps that the Cy5 fluorescence should undergo before DNA unwinding. (H) Shown is a pie chart identifying the percentages of each theoretical model, as predicted by linear combination for (G). To see this figure in color, go online.

Microscope

Objective-type total internal reflection microscopy was conducted using an inverted microscope (ECLIPSE Ti-E; Nikon, Tokyo, Japan) and lasers to excite Cy3 at 532 nm (Compass 215M; Coherent, Santa Clara, CA) and Cy5 at 637 nm (Cube 635-25C; CVI Melles Griot, Albuquerque, NM). The fluorescence signals from the samples were collected by sequential passage through an objective (CFI Apochromat TIRF 100XC Oil, numerical aperture = 1.49; Nikon, Tokyo, Japan), dichroic mirrors to separate fluorescence from Cy3 and Cy5, and barrier filters for Cy3 (FF01-593/40; Semrock, Lake Forest, IL) and Cy5 (670DF40; Omega Optical, Brattleboro, VT) to eliminate the background light. The fluorescence signals in a field of view were filtered (570–615 nm for Cy3 and 650–690 nm for Cy5), simultaneously imaged using a dual-view apparatus, and recorded with an electron multiplying charge-coupled device camera (DU-860; Andor Technology, Belfast, UK). The recorded images were analyzed using Solis software (Andor Technology).

Single-molecule experiments were performed using two laser-excitation methods (9). In the experiments using a nail-polish-sealed flow cell for photobleaching-step analysis in the absence of nucleotides or the presence of ATPγS, the lasers were simultaneously and continuously incident on the sample plane, and the time resolution was 33 ms. For DNA-unwinding assays performed using a flow cell sealed with double-sided tape in the presence of ATP, the lasers were simultaneously incident on the sample plane for 100 ms/s using mechanical shutters (LS3; Uniblitz, Rochester, NY) and Andor IQ software to minimize photobleaching of the dyes; thus, the time resolution was 1 s.

Results

Single-molecule imaging of Cy5-labeled UvrDΔ40C binding to DNA in the absence of nucleotides or the presence of ATPγS

Mechanic et al. proposed the monomer model to explain DNA unwinding by UvrD based on their genetic and biochemical data obtained using UvrDΔ40C (25). To determine whether UvrDΔ40C binds to DNA in only a monomeric form, the number of UvrDΔ40C or UvrD molecules that bind to DNA in the absence of nucleotides was quantified (9). In the experiment, an 18-bp dsDNA with a 20-nt 3′ ssDNA tail (7,33, 34, 35, 36) that was labeled with Cy3 (9) was surface-immobilized via streptavidin-biotin interactions (Fig. 1 A). Then, the number of photobleaching steps was counted for a Cy5-labeled Cys-Ala mutant (Cy5-UvrDΔ40C) that colocalized with the sparsely immobilized Cy3-DNA using a dual-view apparatus. The colocalized Cy5-UvrDΔ40C bound to the DNA, as almost no Cy5-UvrDΔ40C was imaged on the PEGylated glass surface in the absence of the DNA (9). The UvrDΔ40C mutant only contained Cys⁵² (out of six Cys residues present in wild-type UvrD) and, thus, was labeled with a single Cy5 molecule with high specificity and a high labeling ratio of 79%. Buffer U (6 mM NaCl, 2.5 mM MgCl₂, 10% (v/v) glycerol, and 25 mM Tris-HCl (pH 7.5)) was used in all of the experiments in this study (9,36). Fig. 1, B and C show time courses of fluorescence intensities for different fluorescent spots in which one or two different photobleaching events occurred.

The same experiment was performed in the presence of ATPγS, a nonhydrolyzable ATP analog (Fig. 2 A), to examine whether the ligand increased the number of UvrDΔ40C or UvrD molecules that could bind to the DNA (9). Fig. 2, B–D show time courses of the fluorescence intensities of different fluorescent spots in which one, two, or three different photobleaching events occurred.

The mean photobleaching time of Cy5 molecules that were bound to UvrDΔ40C under the experimental condition was ∼20 s (Fig. S1). The mean off-times were 53 ± 2 s (in the absence of nucleotides) and 44 ± 5 s (in the presence of ATPγS) for the first bound UvrDΔ40C, 2.9 ± 0.2 s (in the absence of nucleotides) and 32 ± 7 s (in the presence of ATPγS) for the second bound UvrDΔ40C, and 21 ± 6 s (in the presence of ATPγS) for the third bound UvrDΔ40C (Fig. S2). The observed fluorescence decrease for the first bound UvrDΔ40C was assumed to be caused mostly by photobleaching. It is possible that a fraction of the fluorescence decrease for the first, second, or third bound UvrDΔ40C was caused by the dissociation of Cy5-UvrDΔ40C from the DNA. These were indistinguishable from each other. However, such dissociation that might have occurred during the observation did not affect the quantification of the number of bound proteins.

Multiple UvrDΔ40C molecules bound to DNA in the absence of nucleotides or the presence of ATPγS

Fig. 3, A and B show the distributions of the number of photobleaching steps that were observed in the presence of 2.0 nM Cy5-UvrDΔ40C alone or the presence of both 2.0 nM Cy5-UvrDΔ40C and 1 mM ATPγS in solution, respectively. The distribution for Cy5-UvrDΔ40C alone showed two photobleaching steps, suggesting that two UvrDΔ40C molecules can bind the DNA substrate. In contrast, the distribution observed in the presence of both Cy5-UvrDΔ40C and ATPγS showed a smaller fraction for a single photobleaching step and larger fractions for two or three photobleaching steps. These results suggest that the presence of ATPγS increased the number of UvrDΔ40C molecules that could bind to DNA (7,34) by up to three, which was observed for wild-type UvrD (9). These results demonstrate that at least two and at most three UvrDΔ40C molecules bound the same DNA substrate and that the presence of ATPγS promoted the binding of more UvrDΔ40C molecules to DNA, which was also observed for wild-type UvrD (9). The binding of two UvrDΔ40C molecules was also observed for a duplex DNA substrate with a shorter 3′ ssDNA tail (18-bp DNA with a 12-nt 3′ ssDNA tail) in the absence or in the presence of 1 mM ATPγS (Fig. S6).

Figure 3.

Experimentally obtained and theoretically predicted distributions of the number of photobleaching steps for the Cy5-UvrDΔ40C that bound to an 18-bp DNA with a 20-nt 3′ ssDNA tail. (A and B) Experimentally obtained distributions derived from solutions containing (A) Cy5-UvrDΔ40C alone or (B) Cy5-UvrDΔ40C and 1 mM ATPγS. The total number of analyzed fluorescent spots is indicated for each condition. (C) Shown are theoretical distributions of the number of photobleaching steps for the one-molecule, two-molecule, and three-molecule models. (D and E) Shown are pie charts identifying the percentages of each theoretical model, as predicted by linear combination for (D) Cy5-UvrDΔ40C alone and for (E) Cy5-UvrDΔ40C and 1 mM ATPγS. To see this figure in color, go online.

Fig. 3 C shows the predicted distributions of the number of photobleaching steps for the one-molecule, two-molecule, or three-molecule models of UvrDΔ40C binding to DNA (Supporting Material, Result S1). The predicted distributions were calculated based on the labeling ratio of Cy5-UvrDΔ40C (79%), which indicated that nonlabeled UvrDΔ40C (21%) that bound to DNA was not visualized by single-molecule fluorescence imaging. To estimate the ratios of the one-molecule, two-molecule, or three-molecule UvrDΔ40C-DNA complexes found experimentally, each distribution was fitted based on a linear combination of the predicted distributions. The optimal ratios were obtained by minimizing χ² values, which were the statistics representing the differences between the experimentally obtained distributions and the fitted distributions created using estimated ratios. The distribution for UvrDΔ40C alone, which did not show three photobleaching steps, was fitted based on a linear combination of the one-molecule and two-molecule models. The percentages of the one-molecule and two-molecule models obtained by performing the fitting were 85 and 15%, respectively (Fig. 3 D). The distribution for UvrDΔ40C in the presence of ATPγS, having fractions representing one, two, and three photobleaching steps, was fitted based on a linear combination of the one-molecule, two-molecule, and three-molecule models. The percentages of the monomer, dimer, and trimer models obtained by performing the fitting were 45, 41, and 14%, respectively (Fig. 3 E).

Visualization of a single step or multiple steps by increased Cy5 fluorescence just before DNA unwinding

The results in Fig. 3 show that the presence of ATPγS increased the number of UvrDΔ40C that could bind to DNA. The determined amount of Cy5-UvrDΔ40C that bound to DNA in the presence of an ATP analog indicates that multiple Cy5-UvrDΔ40C molecules are involved in DNA unwinding in the presence of ATP. Next, the number of helicases bound to DNA and driving unwinding in the presence of ATP and the association/dissociation events between the helicase and DNA in the presence of ATP were quantified by simultaneous single-molecule visualization studies using a flow cell (Fig. 4 A; (9)). DNA unwinding was monitored through the disappearance of Cy3 fluorescence, which was attached to one of the two oligonucleotides that formed the dsDNA.

As observed for wild-type UvrD, the Cy3 and Cy5 fluorescence intensities decreased almost simultaneously to their respective background levels in many traces (Fig. 4, B–D), which shows that the UvrDΔ40C unwound the DNA and then immediately dissociated from it (9). Moreover, the Cy5 fluorescence intensity in such traces increased either in multiple steps (Fig. 4, B and C) or in a single step (Fig. 4 D) just before completion of the DNA-unwinding process. Of the 89 traces analyzed, those that exhibited multiple and single steps accounted for 47 and 42 traces, respectively. The 47 traces included three traces that showed three association or dissociation steps.

The mean photobleaching times of Cy5 molecules bound to UvrDΔ40C and of Cy3 bound to DNA under the experimental condition were 250 ± 40 s and >600 s, respectively (Fig. S3). Although fluorescence decrease due to photobleaching was not distinguishable from the fluorescence decrease due to DNA unwinding and UvrDΔ40C dissociation in this assay, photobleaching did not significantly interfere with the analysis, as the Cy3 and Cy5 photobleaching times were long, and only a synchronized decrease in both Cy3 and Cy5 fluorescence signals were considered as positive signals.

DNA unwinding was complete immediately after additional UvrDΔ40C molecule(s) bound to the DNA

As observed with Cy5-labeled wild-type UvrD (9), most traces with multiple steps for Cy5-UvrDΔ40C demonstrated rapid DNA unwinding immediately after additional UvrDΔ40C molecule(s) bound to the DNA. Fig. 4 E shows the distribution of the dwell time between the appearance of the second step and completion of the DNA-unwinding process (τ₂) in traces that showed two-step Cy5 fluorescence increases just before completion of the DNA-unwinding process. The mean dwell time, which was obtained using a single exponential fit, was 2.1 ± 0.1 s. The last increase in Cy5 fluorescence in the traces before DNA unwinding indicates that the three-step Cy5 fluorescence increase or decrease had a mean dwell time of 7.3 ± 3.2 s. The dwell time had a relatively large uncertainty because it was calculated by simple averaging with a limited amount of data (n = 3).

These results suggest that UvrDΔ40C unwinds DNA in the same manner as the wild-type UvrD: a single UvrDΔ40C molecule binds to the DNA, waits for an additional UvrDΔ40C molecule(s) to bind, and completes the DNA-unwinding process. Traces showing a three-step fluorescence change were also observed for wild-type UvrD (9).

Multiple UvrDΔ40C molecules were involved in unwinding the DNA

The traces with multiple steps show that multiple UvrDΔ40C molecules were involved in DNA unwinding. The traces with a single step can be explained using the two-molecule model, even though they seem to support the one-molecule model, because the ratios of the step numbers for the Cy5 fluorescence changes (Fig. 4 F) are better reproduced by the two-molecule model than by the one-molecule model (Fig. 4 G). The ratios for each model were calculated based on the estimated number of Cy5 molecules bound per UvrDΔ40C molecule and the dwell time for the second UvrDΔ40C association (Supporting Material, Result S2). The percentages of the one- and two-step fluorescence changes in the experimentally obtained traces, i.e., 42/(42 + 47) = 47% and 44/(42 + 47) = 49%, were closest to those predicted by the two-molecule model (48 and 52%, respectively). To estimate the ratios of the one-step, two-step, and three-step Cy5 fluorescence increase observed experimentally, the experimentally obtained distribution was fitted based on a linear combination of the predicted models. The percentages of the one-molecule, two-molecule, and three-molecule models obtained by performing the fitting were 1, 92, and 7%, respectively (Fig. 4 H). Therefore, it can be concluded that DNA unwinding was mostly performed not by a single UvrDΔ40C molecule but, rather, by two UvrDΔ40C molecules, and in some cases by three UvrD molecules. The two-molecule model is supported by the result that the total time (time required for UvrDΔ40C association + completion of DNA unwinding) for the two-step Cy5 increase data was comparable to that of the one-step Cy5 increase data (Fig. S7). Two-step and single-step Cy5 fluorescence increase, as well as Cy5 fluorescence increase corresponding to three Cy5 molecules just before DNA unwinding, were also observed for a duplex DNA substrate with a shorter 3′ ssDNA tail (18-bp DNA with a 12-nt 3′ ssDNA tail) (Fig. S6).

Association/dissociation rates for the UvrDΔ40C-DNA interaction

Direct visualization of Cy5-labeled UvrDΔ40C can be used to determine some of the kinetic rate constants for the UvrDΔ40C-DNA interaction. To determine the approximate values of the rate constants, the Cy5 fluorescence increase/decrease steps were interpreted as UvrDΔ40C association/dissociation events. This approximate determination is possible because almost 80% of the UvrDΔ40C molecules were labeled with Cy5. The traces showed various intensities for one-, two-, and three-step Cy5 fluorescence increases or decreases, which was indicative of sequential UvrDΔ40C monomer association/dissociation. Some traces seemed to show simultaneous binding of UvrDΔ40C dimers to DNA or dissociation of UvrDΔ40C dimers from DNA; however, the binding of preassembled dimers to DNA and the dissociation of dimers from DNA could not be distinguished from sequential binding of two UvrDΔ40C monomers and sequential dissociation of two UvrDΔ40C monomers, respectively, because of the lack of temporal resolution (Supporting Material, Result S3). The two- and three-step increases or decreases in the Cy5 fluorescence intensity before or after association/dissociation were used as criteria for confirming the steps representing association/dissociation of the two and three UvrDΔ40C molecules, respectively. Therefore, the kinetic scheme of UvrD-DNA interactions proposed in the previous study for wild-type UvrD (9) can be used for UvrDΔ40C (Fig. 5 A).

Figure 5.

Association and dissociation rate constants. (A) Shown is a kinetic scheme of the UvrD-DNA interaction. This scheme was conceived based on the assumption that the number of UvrD molecules involved in the interaction is at most three (9). The UvrD molecule initially forms a complex with DNA in the form of UD, U₂D, or U₃D; undergoes a series of UvrD association/dissociation events; and then forms the UvrD oligomer. The oligomer isomerizes on the DNA to form an active complex (U₂D^∗ or U₃D^∗) that unwinds the DNA. (B) Shown are dwell-time distributions of the indicated states in which up to two UvrD monomers were involved. Each distribution fit satisfactorily with a single exponential, which yielded the corresponding association and dissociation rate constants. k₁ and k₂ are association rate constants under 2 nM UvrDΔ40C concentration. The distributions do not include the dwell time for association and dissociation process during DNA unwinding. (C) Shown is a comparison of the obtained rate constants between UvrDΔ40C and wild-type UvrD. K_on(obs) are the observed association rate constants under a 2 nM UvrDΔ40C concentration. The error bars represent the standard errors. To see this figure in color, go online.

Fig. 5 B shows the dwell-time distributions of the indicated states in the association and dissociation reactions between UvrDΔ40C molecules and DNA in the DNA-unwinding experiments, which involved up to two UvrDΔ40C molecules. The distributions did not include a dwell time for the dissociation occurring just before completion of the DNA-unwinding process. Each distribution fit satisfactorily with a single exponential, which yielded the corresponding mean dwell time. As with wild-type UvrD (9), the dwell times for both the dissociation and association processes decreased as the number of UvrDΔ40C molecules involved increased (Fig. 5 C).

Comparison of the rate constants between UvrDΔ40C and wild-type UvrD showed comparable association and dissociation rates for the first UvrD and that the rate of the second UvrD was ∼2.5-fold higher for UvrDΔ40C.

A comparison of the equilibrium constants that were calculated using the relationship K_n = k_-n/(k_n × 2 × 10⁻⁹) (n = 1, 2) (mean ± standard error) is shown in Table S4. Comparison of the K₁ and K₂ values indicates that first bound and second bound UvrDΔ40C molecules had similar DNA affinity (K₁ = [1.3 ± 0.0] × 10⁸ M⁻¹ and K₂ = [1.2 ± 0.2] × 10⁸ M⁻¹) to the first bound and second bound wild-type UvrD molecules: K₁ = (1.6 ± 0.2) × 10⁸ M⁻¹ and K₂ = (1.3 ± 0.2) × 10⁸ M⁻¹.

Discussion

Number of UvrDΔ40C molecules bound to DNA in the absence of nucleotides or the presence of ATPγS

The photobleaching steps observed in single-molecule fluorescence images were analyzed to determine the number of UvrDΔ40C molecules that bound to DNA substrates with a 20-nt 3′ ssDNA tail in the absence of nucleotides. The analysis revealed that at most two UvrDΔ40C molecules bound to DNA (Fig. 3 A), which was fewer than the three wild-type UvrD molecules bound to DNA, as determined by single-molecule direct visualization (9) and analytical ultracentrifugation at saturating protein concentration (34). Three wild-type UvrD molecules bound to DNA at the same protein concentration, as determined by goodness-of-fit tests that best fitted a trimer model with a minimal χ² value (9). For UvrDΔ40C, that number seemed to be one in most cases (85%, Fig. 3 D), as determined by fitting the photobleaching-step histogram with a linear combination of the one-molecule and two-molecule models. UvrDΔ40C, which partially lacks the C-terminal amino acids of wild-type UvrD, was likely to directly self-interact on ssDNA and form an oligomer, as wild-type UvrD molecules were reported to have a low affinity for blunt duplex DNA (9) and a lower limit of the estimated wild-type UvrD site size on poly(dT) of 10 ± 2 nt (37). Further, two UvrDΔ40C molecules were able to bind dsDNA with a shorter 12-nt 3′ ssDNA tail (Fig. S6). Lee et al. used DNA substrates with a 7-nt ssDNA tail and obtained structures of UvrDΔ40C monomer that bound to the DNA by x-ray crystallography (21), indicating that the UvrDΔ40C monomer can bind to an ssDNA-dsDNA junction with 7-nt ssDNA, but two UvrDΔ40C monomers cannot. Maluf et al. reported that the wild-type UvrD monomer bound tightly to dsDNA with an ssDNA (≥4 nt) (7).

The results obtained in this study may challenge the notion that UvrDΔ40C fails to dimerize. The assembled state of UvrDΔ40C in the absence of nucleotides under 100 or 200 mM NaCl conditions was studied by Mechanic et al. using biochemical methods (25). The buffer conditions are listed in Table S5. The authors examined the assembled state of UvrDΔ40C by analytical-sedimentation-equilibrium-ultracentrifugation experiments ([UvrDΔ40C] = 2.8, 5.0, and 8.4 μM) and concluded that UvrDΔ40C failed to dimerize, as all data for UvrDΔ40C were most consistent with a model for a single species with a monomeric molecular mass (25). It is unknown how the presence of DNA affects the sedimentation coefficients of UvrDΔ40C in the absence of nucleotides because the authors did not show results for the sedimentation experiment in the absence of nucleotides but in the presence of DNA. The same group also performed high-pressure liquid chromatography gel filtration ([UvrDΔ40C] = 5.1 μM when injected into the column) to determine the apparent molecular mass of UvrDΔ40C. They reported that elution of UvrDΔ40C resulted in a single peak, whereas wild-type UvrD eluted as a single peak that was broader than the peak for UvrDΔ40C, with an apparent molecular mass between that expected for a monomeric and dimeric proteins, as observed by Runyon et al. (37). The authors concluded that the eluted UvrDΔ40C was homogeneous and contained only UvrDΔ40C monomers.

However, Maluf et al. claimed that UvrDΔ40C might form a dimer in a buffer containing lower salt concentrations, which supports the results in this study. They showed that UvrDΔ73C was still capable of forming a dimer in a buffer containing 20 mM NaCl, but not in 200 mM NaCl, using sedimentation-equilibrium experiments (34). Nonetheless, the dimerization equilibrium constant for UvrDΔ73C was 25 times smaller than that of wild-type UvrD.

The presence of 1 mM ATPγS increased the number of UvrDΔ40C molecules that bound to DNA (three at most), which is similar to the results obtained using single-molecule direct visualization of wild-type UvrD (9). For wild-type UvrD, the photobleaching-step distribution in the presence of the ligand shifted to a larger value compared with that in the absence of the ligand, but it was observed that three UvrD molecules bound to DNA under the same protein-concentration condition, which was obtained by goodness-of-fit tests that best fitted a three-molecule model with a minimal χ² value. In contrast, for UvrDΔ40C, the number seemed to be either one (45%) or two (41%) in most cases (Fig. 3 E), which was determined by fitting the photobleaching-step histogram using a linear combination of the one-molecule, two-molecule, and three-molecule models.

The effects of nucleotide-binding on the UvrD conformation were previously examined using trypsin and chymotrypsin as probes (38). It was reported that the presence of nucleotides (ATP or ATPγS) or ssDNA (M13mp11 ssDNA) stabilized the 72-kDa tryptic polypeptides against further cleavage, indicating that either the UvrD conformation was altered or that the cleavage site was protected by occlusion. This result demonstrates that the wild-type UvrD in the presence of DNA and ATPγS assumed a similar conformation to that obtained in the presence of DNA and ATP. A recent intramolecular single-molecule Förster resonance energy transfer study showed that ATPγS altered conformation of the 2B subdomain of the wild-type UvrD monomer (12). In that study, it was reported that the 2B subdomain can populate at least four rotational conformational states, which were referred to as S1, S2, S3, and S4, from most open to most closed conformational states, and that the addition of ATPγS slightly changed the population distribution of the four rotational conformational states; the addition increased the S3 population while decreasing the S4 population. They also reported that including ATPγS with a wild-type UvrD dimer-DNA complex further increased the S3 population at the expense of the S4 and S2 states. They concluded that including ATPγS increased the populations of the more closed states relative to those of the more open states. Taken together with the results that ATPγS increased the number of UvrDΔ40C and wild-type UvrD molecules that could bind to DNA, these findings indicate that more closed conformational states induced by ATPγS promote UvrD dimer-DNA complex formation.

Mechanic et al. performed analytical-sedimentation-velocity-ultracentrifugation experiments and showed that the presence of ATPγS altered the UvrDΔ40C conformation (25). They reported sedimentation coefficients for UvrDΔ40C in the absence of ligands, the presence of nucleotides, and the presence of DNA and nucleotides. The sedimentation coefficient (s_20,w) for UvrDΔ40C decreased in the presence of an ATP analog (adenylyl-imidodiphosphate (AMP-PNP)) relative to that of the protein alone (from 7.0 ± 0.3 to 5.7 ± 0.3, respectively) and decreased further to 4.8 ± 0.1 in the presence of (dT)₁₆ and AMP-PNP.

Using high-pressure liquid chromatography gel filtration, the same group reported that the apparent molecular mass of UvrDΔ40C slightly increased in the presence of ATP. They also performed gel filtration using a preformed enzyme-ssDNA complex and compared the apparent molecular mass of UvrDΔ40C in the absence and presence of ATP and oligonucleotide (dT)₁₆. The authors concluded that UvrDΔ40C did not dimerize because the apparent molecular mass increased by approximately the same amount as when only ATP was present.

The discrepancy between the conclusion of this study that multiple UvrDΔ40C molecules are bound to DNA and might form an oligomer and the conclusion by Mechanic et al. that UvrDΔ40C is monomeric regardless of the presence or absence of DNA is probably derived from the lack of data using dsDNA with a 3′ ssDNA tail and the shorter temporal interaction of the second UvrDΔ40C for DNA than wild-type UvrD. A previous report by Maluf et al. (7) and a previous study by the author’s group (9) showed that wild-type UvrD had a high affinity to the ssDNA/dsDNA junction. These properties could have made it difficult to detect binding of multiple UvrDΔ40C molecules on DNA composed of only an ssDNA in previous biochemical studies. The photobleaching distribution for UvrDΔ40C alone (Fig. 3 A) showed a small fraction for two photobleaching steps, which demonstrates that two UvrDΔ40C molecules were rarely bound to the DNA substrate (Fig. 3 D). The photobleaching distribution for UvrDΔ40C in the presence of ATPγS (Fig. 3 B) showed higher fractions for two or three photobleaching steps than did the distribution for UvrDΔ40C alone (Fig. 3 E). Analysis of the kinetic rate constants of the UvrDΔ40C-DNA interaction shows that the association and dissociation rates of the first bound UvrD were comparable and that the rates of the second bound UvrD were ∼2.5-fold higher for UvrDΔ40C, which suggests that UvrDΔ40C oligomerization on DNA was transient.

Multiple protein association to DNA and the unwinding activity of UvrDΔ40C

In this study, the association of two or three UvrDΔ40C molecules to DNA was observed just before completion of DNA unwinding in the presence of ATP; thus, the multiple UvrDΔ40C molecules appeared to be directly related to the DNA-unwinding activity of UvrDΔ40C. This multiple-molecule association was observed for wild-type UvrD in previous biochemical and single-molecule studies that have suggested that wild-type UvrD has optimal activity in its oligomeric form (7, 8, 9, 10, 11).

The higher percentage of the one-step fluorescence change and the lower percentage of the two-step fluorescence change in the experimentally obtained traces likely reflect the short dwell time (mean τ₂ was 2.1 ± 0.1 s). The temporal resolution of the unwinding experiments was 1 s; thus, these experiments did not detect all of the two-step Cy5 fluorescence increase just before completion of the DNA-unwinding process.

Data from a recent study using single-molecule force-fluorescence microscopy, which enabled simultaneous DNA-unwinding activity measurement with optical tweezers that exerted tension on the DNA and counting of helicase binding based on fluorescence detection, indicated that wild-type UvrD dimers are capable of long-distance unwinding (11), consistent with previous reports (7,35). The study also demonstrated that wild-type UvrD monomers showed long-distance unwinding under tension (13 pN), but in the absence of tension, wild-type UvrD monomers should not be competent to unwind 20-bp dsDNA, which was observed in a previous biochemical study (7) and with UvrDΔ40C in this study.

No reports have described a single-molecule analysis of the assembly states of UvrDΔ40C in the presence of DNA and ATP; Mechanic et al. examined the characteristics of UvrDΔ40C by biochemical assays using buffers containing 20 mM NaCl (Table S5) and compared that of wild-type UvrD (25,30). The authors of those studies reported that UvrDΔ40C had similar ssDNA-binding abilities and turnover rates for ssDNA-stimulated ATP binding and hydrolysis and dsDNA-unwinding efficiency when compared with wild-type UvrD. However, these experiments were performed under very different solution conditions (an NaCl concentration (20 mM) and glycerol concentrations (0 or 10%)) (Table S5) that were used for analytical-sedimentation-equilibrium-ultracentrifugation and high-pressure liquid chromatography gel filtration experiments (NaCl concentrations (100 or 200 mM) and a glycerol concentration (20%)), which concluded that UvrDΔ40C is active as a monomer. Maluf et al. claimed that UvrDΔ40C may form a dimer while unwinding DNA, as the above experiments by Mechanic et al. were performed in a solution with a low NaCl concentration (20 mM), in the absence of glycerol (0%), at a low pH (7.5), and at a high temperature (37°C), which were favorable for oligomer formation.

Thus, there is a gap between the conclusions obtained using high- and low-NaCl solutions. The buffer U, used in this study, contained 6 mM NaCl, which was far less than that in the physiological condition and different from the buffer conditions used for analytical-sedimentation-equilibrium-ultracentrifugation and high-pressure liquid chromatography gel filtration experiments by Mechanic et al. (25). Then, another single-molecule visualization was performed under a physiologically relevant buffer condition (200 mM NaCl). This buffer condition was similar to the one used in velocity-ultracentrifugation experiments by Mechanic et al. (25), except for the higher NaCl concentration (200 mM), which was used in their analytical-sedimentation-equilibrium-ultracentrifugation experiments (Table S5) and was reported to be more unfavorable for dimer formation. Despite such a high salt concentration, two-step fluorescence increase of Cy5 bound to UvrDΔ40C was observed, indicating that two UvrDΔ40C molecules can bind to the DNA. Moreover, Cy5 fluorescence increases, with a two-or-more-fold increase in intensity compared with those of the Cy5 fluorescence intensity increases from one Cy5-UvrDΔ40C, were observed just before completion of the DNA-unwinding process (Fig. S8). These results suggest that multiple UvrDΔ40C molecules also participate in DNA unwinding under the physiologically relevant salt condition.

Fewer UvrDΔ40C association events occurred under the 200 mM NaCl buffer condition compared with that in the 6 mM NaCl condition. The UvrD concentration in vivo was predicted to range from 1.4 to 4 μM (1000–3000 molecules/cell) (39), and the effective concentration of a nick, which is created by other DNA-repair enzymes near a damaged or mismatch DNA, in the genomic DNA in vivo is ∼1 nM, which is much higher than the estimated concentration of dsDNA-ssDNA junction in the single-molecule assays (∼3 pM). Thus, the rare UvrDΔ40C association events observed by the single-molecule visualization must be much more frequently performed in vivo. On the other hand, the UvrDΔ40C/dsDNA-ssDNA junction in the single-molecule assays was ∼1 × 10³, which is comparable to the UvrD/nick in vivo (∼1 × 10³). The comparable ratio of UvrD/nick together with the higher UvrD and nick concentrations in vivo suggests that association of multiple UvrDΔ40C molecules to DNA and their participation in DNA unwinding observed under the 200 mM NaCl condition is relevant to UvrD function in vivo, though an in vivo environment, including high-crowding conditions, would somehow modulate dimerization of UvrD on DNA.

Kinetic mechanism of the UvrDΔ40C-DNA interaction

The previous study by the author’s group examined the association/dissociation dynamics of wild-type UvrD in the presence of ATP (9). In this study, the same two-color visualization approach was used, which enabled the determination of some approximate kinetic rate constants for the UvrDΔ40C-DNA interaction. The association/dissociation rate constants obtained in this study (Fig. 5 B) were on the same orders of magnitude as those obtained in the past study by the author’s group for wild-type UvrD (Fig. 5 C).

The rate constants in this study have similar internal relationships to those determined in the past study by the author’s group (9) and by Maluf et al., who used double-mixing quenched-flow experiments (35) for wild-type UvrD. The dwell times of both the dissociation and association processes decreased as the number of UvrD molecules involved increased (k₂ was larger than k₁, and k_-2 was larger than k_-1).

Comparison of the rate constants for the association of the first monomer ((D→UD); k_1(wt) = 0.063 ± 0.03 s⁻¹, k_1(Δ40C) = 0.056 ± 0.001 s⁻¹) and the dissociation of a monomer ((UD→D); k_-1(wt) = 0.20 ± 0.02 s⁻¹, k_-1(Δ40C) = 0.22 ± 0.00 s⁻¹) indicate that the rates were comparable and that the first UvrDΔ40C bound had a similar DNA affinity to the first wild-type UvrD bound. This result agrees with the data reported by Mechanic et al. that UvrDΔ40C had a slightly higher dissociation constant for ssDNA than wild-type UvrD in the absence and presence of 1 mM AMP-PNP (30). Comparison of the rate constants for the association of a second bound monomer ((UD→U₂D); k_2(wt) = 0.10 ± 0.00 s⁻¹, k_2(Δ40C) = 0.27 ± 0.03 s⁻¹) and dissociation of the second bound monomer ((U₂D→UD); k_-2(wt) = 0.40 ± 0.06 s⁻¹, k_-2(Δ40C) = 1.1 ± 0.1 s⁻¹) indicate that both rates for UvrDΔ40C were ∼2.5-fold larger than those for wild-type UvrD and that the second bound UvrDΔ40C monomer attached to and detached from DNA more frequently than the second bound wild-type UvrD monomer.

The higher characteristics of both association and dissociation rate constants were canceled out by division when calculating the equilibrium constant (Table S4). It was observed that 1 mM ATPγS seemed to change conformation of UvrDΔ40C and slightly increased the number of UvrDΔ40C molecules that bound to DNA (Fig. 3 B).

Traces in which a UvrDΔ40C preassembled dimer seemed to associate with DNA and to dissociate from DNA were found, and their corresponding association/dissociation rates constants were determined (Fig. S5). Comparisons of the rate constants for the sequential association of two UvrDΔ40C monomers (D→UD and UD→U₂D) with the association of a seemingly preassembled UvrDΔ40C dimer (D→U₂D) and comparison of the rate constants for the sequential dissociation of two UvrDΔ40C monomers (U₂D→UD and UD→D) with the dissociation of a UvrDΔ40C dimer (U₂D→D) indicate that association of the apparently preassembled UvrDΔ40C dimer and dissociation of the apparently UvrDΔ40C dimer were indistinguishable from UvrDΔ40C dimer formation through sequential DNA binding of two UvrDΔ40C monomers and sequential dissociation of two UvrDΔ40C monomers from DNA, respectively (Supporting Material, Result S3). Such preassembled UvrDΔ40C dimer association and UvrDΔ40C dimer dissociation appeared to be detected in this study; however, it is difficult to definitively conclude that detection occurred because of the lack of temporal resolution.

UvrDΔ40C-DNA interaction just before completion of the DNA-unwinding process

In the presence of ATP, two or three UvrDΔ40C molecules that bound to DNA just before completion of the DNA-unwinding process were likely to directly self-interact on ssDNA and form an oligomer, as discussed earlier. The self-interaction and oligomer formation in the presence of ATP was originally proposed from a nonlinear sigmoidal dependence of DNA-unwinding efficiency on [wild-type UvrD]/[DNA] (7) and is supported by the results of DNA-unwinding experiments for UvrDΔ40C molecules using 18-bp DNA with a 12-nt 3′ ssDNA tail (Fig. S6). The 3′ ssDNA length (12 nt) was reported to be the minimal requirement to complete unwinding of the duplex portion of the DNA (7). Although this DNA substrate having a shorter 3′ ssDNA tail than the 18-bp DNA with a 20-nt 3′ ssDNA tail was considered to accommodate fewer UvrDΔ40C molecules (9,33), the experiments revealed that two or three UvrDΔ40C molecules were bound even to the DNA just before DNA unwinding. Nevertheless, it was very difficult to provide absolute proof of the involvement of the self-interaction or oligomerization in DNA unwinding.

The conflicting “independent monomer” model has been proposed for some helicases, such as Dda (40) and RecQ (41), which says that these helicases can unwind DNA without their self-interaction or oligomerization: multiple helicases can align along DNA, move independently toward the same direction, stabilize denatured states of DNA, and finally perform efficient DNA unwinding. Note that this model can incorporate some interaction between helicases. UvrD might also use this independent helicase activity; however, Maluf et al. reported that DNA unwinding by UvrD could not be explained by the model (7). Moreover, recent findings showed that UvrD dimerization, the binding of a second UvrD to a first bound UvrD, shifted the 2B conformation of the first bound UvrD to a more closed state, resulting in the activation of helicase activity (12). Thus, it is probable that multiple UvrDΔ40C molecules that bound to DNA just before completion of the DNA-unwinding process somehow interact with each other.

If the assumption is made that UvrDΔ40C can form an oligomer on DNA, three kinetic steps are thought to occur during the mean dwell time of multiple UvrDΔ40C-bound states (2.1 ± 0.1 for two UvrDΔ40C-bound states or 7.3 ± 3.2 s for three UvrDΔ40C-bound states) before completion of the DNA-unwinding process (Fig. 4 E). These include 1) the translocation of late-coming UvrDΔ40C molecule(s) that was (were) bound to the ssDNA to form a UvrDΔ40C oligomer with those that were bound previously, 2) the isomerization of a nonproductive oligomer to become ready for unwinding the DNA, and 3) the unwinding of the DNA. UvrDΔ40C shares similar biochemical characteristics with wild-type UvrD (25,30), and thus, late-coming UvrDΔ40C might require almost the same time for ssDNA translocation as wild-type UvrD to form a UvrDΔ40C oligomer with those that were bound previously and to promote DNA unwinding. The late-coming UvrDΔ40C monomer thus can interact with the prebound UvrDΔ40C monomer(s) through direct contact or through translocation along the 3′ ssDNA tail in less than 0.1 s because wild-type UvrD monomers are known to perform ATP-dependent translocation along ssDNA with a biased 3′-to-5′ directionality at a translocation rate of ∼190 nt/s, a processivity of 769 ± 1 nt (42), and a translocation rate along dT of ∼190 nt s⁻¹ (43). In addition, a functional UvrDΔ40C oligomer must completely unwind the 18-bp dsDNA substrate in <0.5 s without dissociating as the unwinding rate, and the processivity of wild-type UvrD is reported to be 68 ± 9 bp/s (7) and is 40–50 bp (44), respectively. In summary, the time required for the late-coming UvrDΔ40C monomer to find the prebound UvrDΔ40C molecule(s) and unwind the 18-bp dsDNA would be less than 0.5 s.

Intriguingly, the mean dwell time between the appearance of the second step and the completion of the DNA-unwinding process (τ₂) for UvrDΔ40C (2.1 ± 0.1 s) was shorter than that reported for wild-type UvrD (2.7 ± 0.2 s) (9). Therefore, if the assumption is made that UvrDΔ40C can form an oligomer on DNA, time consumed for the isomerization process for UvrDΔ40C would be shorter than that for wild-type UvrD. This shorter dwell time agrees with the statement by Mechanic et al. that the rate of unwinding catalyzed by UvrDΔ40C molecules was reproducibly slightly greater than that of wild-type UvrD molecules (25). The C-terminal 40-aa deletion should somehow modulate the isomerization process. Of note, the second bound UvrDΔ40C exhibited more frequent association/dissociation with/from DNA, but its equilibrium constant was similar to that for wild-type UvrD (Fig. 4 B).

As mentioned above in the Discussion, the single-molecule imaging assays performed in this study could not distinguish the association of a preassembled UvrD dimer from the sequential binding of two UvrD monomers (Supporting Material, Result S3). Sequential binding can be mistaken for preassembled binding because of the temporal resolution limit of the experiment. The apparent preassembled UvrDΔ40C dimer was also found just before completion of the DNA-unwinding process, indicating that a preassembled UvrDΔ40C dimer can unwind the DNA (10). The mean dwell time of the seemingly preassembled dimer just before completion of the DNA-unwinding process (τ_2→0(Δ40C)) was = 1.2 ± 0.1 s, which was shorter than the mean dwell time between the appearance of second step and the completion of the DNA-unwinding process (τ_2(Δ40C) = 2.1 ± 0.1 s, Fig. 4 E). In addition, the mean dwell time of the seemingly preassembled dimer just before completion of the DNA-unwinding process for wild-type UvrD (τ_{2→0 (wt)}) was = 2.1 ± 0.1 s (Fig. S5), which was again shorter than the corresponding mean dwell time (2.7 ± 0.2 s) of wild-type UvrD. Maluf et al. claimed that a wild-type UvrD dimer is a functional helicase that can unwind DNA in less than 50 ms without undergoing an isomerization process on the DNA (35). Our single-molecule imaging assays cannot be used to determine whether the preassembled UvrDΔ40C dimer can unwind DNA without the isomerization process because of the the lack of time resolution. Nonetheless, the time required for isomerization may be shorter than that of wild-type UvrD.

The single-molecule imaging assays performed in this study revealed that multiple UvrDΔ40C molecules were involved in DNA unwinding. Although these results strongly indicate that multiple UvrDΔ40C molecules are necessary to unwind DNA, it cannot completely rule out the possibility that UvrDΔ40C monomer can complete the DNA unwinding. Traces with a single step just before the DNA-unwinding process (Figs. 4 D and S8, C and D) might be explained by the one-molecule model. Therefore, it can be concluded that most DNA unwinding is performed by two or more UvrDΔ40C molecules.

Deletion of C-terminal amino acids decreased the number of helicases bound to DNA and shortened the dwell time of the two-molecule bound state just before completion of the DNA-unwinding process. Previous reports demonstrated the effects of longer C-terminal amino acid deletions on helicase activities (25,29,30); UvrDΔ73C (UvrD_1–647) retained ATPase and DNA-unwinding abilities, whereas UvrDΔ102C lost them. These results indicate that the unstructured C-terminal amino acids (645–720 aa) are dispensable for DNA unwinding and that further deletion of the C-terminal amino acids, which partially consist of the 2A domain containing a conserved DNA-binding motif, abolishes the activity. The DNA-unwinding ability may directly correlate with the length of the C-terminal amino acid deletion. Maluf et al. claimed that UvrDΔ40C may form a dimer because they showed that UvrDΔ73C was still capable of forming a dimer (34).

Data from a recent study involving single-molecule force-fluorescence microscopy demonstrated that UvrD monomers could unwind a limited amount of DNA under tension with low processivity (11). The same study showed that two UvrD conformational states, termed the “closed” and “open” states, arising through an ∼160° rotation of the 2B subdomain about a hinge region connecting it to the 2A subdomain (11,21,22), correlated with movement toward or away from the DNA fork (11). It has been proposed based on crystal structures that the 2B subdomain plays a catalytic role in DNA unwinding (21,23,24). Similar conformational changes were observed for PcrA (23) and Rep (24). Intramolecular cross-linking of a PcrA or Rep monomer into the closed form increases its unwinding processivity (>1000 bp) (45). However, removal of the 2B subdomain in Rep to form RepΔ2B activated the helicase activity of the monomer (46, 47, 48). Therefore, the 2B subdomain was autoinhibitory for Rep monomer helicase activity and thus played a regulatory rather than a catalytic role (47).

The C-terminus plays essential roles in nucleic acid binding for many proteins with helicase and dimerization activities, including the gpα helicase primase from bacteriophage P4, the Rad25 helicase from yeast (49,50), and E. coli RNA helicase CsdA (17). The C-terminal region served as a bipartite extension from the dimeric structure and an RNA-binding module for Geobacillus stearothermophilus CshA (18) and Thermus thermophilus Hera (20). Moreover, it has been found that some mutations of Werner’s syndrome protein are C-terminal truncations outside the conserved helicase motifs (51). The C-terminal domain of Werner’s syndrome protein plays several functions, including binding to dsDNA with a ssDNA tail and mediating the interaction with other proteins (52). These results imply that a region close to the C- terminus of helicases is involved in binding to ssDNA.

In this study, the interrelationships among the C-terminal amino acids, the number of proteins that bound to DNA, and unwinding activity of the nonhexameric helicase UvrD were addressed. In the future, the mechanism whereby the C-terminal amino acids participate in the conformational change and activation of the first bound UvrD will be explored using a more severe form of deletion mutants. Interactions between the C-terminus of UvrD and other DNA-repair proteins such as UvrB and UvrC have been investigated using a C-terminus-truncated UvrD (29). In addition, interactions of UvrD with other DNA-repair proteins, such as MutL (53, 54, 55), UvrAB (56), and RNA polymerase (57,58), may alter its self-interaction and helicase activity as well.

Author Contributions

H.Y. designed and performed the single-molecule imaging experiments, analyzed the data, and wrote the manuscript.

Editor: Keir Neuman.